doc-coauthoring

Guides collaborative document drafting through context gathering, iterative refinement, and reader testing.

Updated May 24, 2026
One-click install
npx skills add https://github.com/MWest2020/skill-forge --skill doc-coauthoring-mwest2020
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doc-coauthoring
Source: https://github.com/MWest2020/skill-forge/tree/main/skills/doc-coauthoring
Command: npx skills add https://github.com/MWest2020/skill-forge --skill doc-coauthoring-mwest2020

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Writing documentation, proposals, and specs alone often produces docs that make sense to the author but confuse readers. This Skill structures the co-authoring process so context is fully transferred, each section is deliberately refined, and the finished doc is verified against a fresh reader before it ships. ## Core Features & Use Cases - Three-Stage Workflow: Context Gathering (info dumps plus clarifying questions), Refinement & Structure (section-by-section brainstorming, curation, and surgical edits), and Reader Testing (cold-reader validation with a fresh agent). - Environment-Aware Tooling: Uses artifacts (create_file/str_replace) on claude.ai or file tools (Write/Edit/Read) in Claude Code, and MCP connectors to pull context from Slack, Google Drive, or SharePoint. - Blind-Spot Detection: Spawns a subagent with only the document text to answer predicted reader questions and flag ambiguity, false assumptions, and contradictions. - Use Case: Draft a decision doc for a migration plan — dump your context, curate brainstormed options per section, then let a fresh Claude instance verify the doc stands on its own before sharing it with the team. ## Quick Start Ask the assistant to guide you through the doc co-authoring workflow to draft a design doc, starting with context gathering.

Frequently Asked Questions about doc-coauthoring

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I co-author a design doc or RFC with Claude?

Start the doc co-authoring workflow, which walks through three stages: context gathering with clarifying questions, section-by-section drafting with brainstormed options you curate, and reader testing with a fresh Claude instance to catch blind spots.

How does the reader testing stage work?

The workflow predicts 5-10 questions real readers would ask, then spawns a subagent containing only the document text and one question. Because the subagent has no conversation context, wrong answers reveal gaps, ambiguity, or contradictions to fix.

Can it pull context from Slack or Google Drive?

Yes, if MCP connectors are available it uses the matching connector tool to fetch channels or documents. Without connectors, you paste content directly, or use WebFetch for publicly readable links in CLI environments.

Does it work in both claude.ai and Claude Code?

Yes. On claude.ai it creates an artifact with create_file and edits via str_replace. In Claude Code or CLI it uses Write once to create the file, Edit for every change, and Read for review passes.

When should I skip this workflow and write freeform?

Skip it for short or low-stakes writing where a three-stage process adds overhead. The workflow is designed for substantial documents like PRDs, decision docs, and specs where reader comprehension matters.